I saw AI was the next internet,
and I went all in.

I'll keep this plain. No jargon, no posturing, and no claim that I'm sharper than anyone on your side. I went all in on AI and started a company on a simple idea: build custom tools for whatever a niche actually needs, using nothing but AI. A regulated shop was just the first real test of that, and it held. I'd rather show you the work and let you push on it.

  • Built a regulated aerospace shop an offline tool that cut estimating work about 60%, plus an AI support layer it reaches from outside its network. No AI runs inside the shop. All through Claude Code.
  • An AI hiring judge that has to cite real evidence, plus a verifier that catches it bluffing.
  • A spread of other custom tools, shipped solo and full stack. More in the range below.

How I think and what I built. Not a plan for your company I have no business writing yet.

I built the tool that did my old job. They kept me on because the value was undeniable: what one person ships with AI is worth far more than the seat costs.

~60% less estimating work 2 → 1 estimators
Live since early 2026, no reported failures
CMMC · ITAR · AS9100 · NADCAP compliant

I started as an estimator. Then I built the estimating tool the shop runs today, end to end through Claude Code on a heavy spine of docs and specs. Here's a short walkthrough, and the real thing to poke at.

When I say regulated, I mean the paperwork holds up.

The customer-facing compliance documents that ship with the tool. Tap to read in place.

Pen-test results go to vetted procurement as a summary, not posted in full on a public page. Same judgment the whole tool is built on.

Two stories. I keep them separate, because the difference is the point.

The tool: offline software
  • No AI inside it. The speed win is software, not a model.
  • Runs fully offline, no telemetry.
  • That offline design is exactly why a regulated shop can run it.
The only AI: a support layer, from outside
  • Email and portal agents. Never installed in the shop.
  • Built off-network, against aliased data, public vendor docs.
  • Nothing AI ever touches the controlled environment, by design.

So how does a shop that can't host AI get AI help? On the far side of a boundary they already control.

  • Multi-channel support, a specialized agent per function, with deliberate scope separation between them.
  • Tier-aware routing, a human-in-the-loop confirmation and escalation layer throughout.
  • Email and portal only. Nothing AI installed inside. An on-site agent add-on is roadmap, not shipped.

Pricing, kept separate: regulated license is $10K flat, unlimited seats. Off-the-shelf is ~$4,999/seat + $99/mo. One active license; a second on the table with a PE owner, not closed.

Don't take a savings number from me. Compute your own.

I built MFC's site and this model. Its rule: never claim a flat savings stat. Start with what an estimator costs you.

40 to 60 percent is typical. A modeled assumption, not a promise.
Year-one net$0
Annual labor savings$0
Year-one investment$0
Payback period-
3-year net$0
Real MFC list prices ($4,999.99 first license, $3,499.99 each additional). Time-reduction is your own assumption. Accuracy gains are real but uncounted here, so this runs conservative. Your numbers against real prices, not a claim about other shops.

A non-coder edits the pricing engine in plain English. Nothing writes until a human says so.

Type a change in plain English. It runs the real settings-editor AI, which proposes a validated diff against the live config so you see exactly what moves. Nothing is written until you Apply.

settings.json · processes & feesDemo · synthetic config

The text box is live: it calls the real settings-editor AI, the same prompt and validation rules the product uses, on a synthetic config. The example buttons are instant and run no model. Either way the shape is the same: the AI proposes, you review a validated diff in place, you own every write.

Nick built a Devil's Advocate GPT. I built one that has to prove every claim it makes.

Screen drops a candidate into one messy business problem and watches how they solve it with AI: a blank build assistant, simulated stakeholders who each hold part of the picture, a gated expert you only reach if you ask, and a deliberate trap. Take it yourself.

What the judge does
  • Scores against an anchored rubric, per dimension.
  • Must quote verbatim evidence for every score.
  • A verifier checks every quote, and catches fabricated evidence.
The honest floor
  • No accuracy percentage. I haven't run a formal study.
  • Reliability is a method, not a number.
  • A Stage-0 prototype, no paying customers.

Admin view Inside the demo, the star button opens the reviewer view: the real judge run on three seeded runs, end to end.

  • Per-dimension scores and the final band.
  • The trap: caught, missed, or pre-empted.
  • Citations: checked, fabricated, flagged.

Synthetic candidates, real scoring. The reviewer view runs the same anchored rubric and citation verifier as the live screen, and shows only genuine judge output, never hand-authored numbers.

Your Law 5 as running code: confident incorrectness comes standard, so fact-check often. The verifier literally fact-checks the grader.

The anchors aren't flukes. The spread:

One person, full stack
Offline tool, Worker backend, database, marketing site, admin CRM, Stripe live, all orchestrated through Claude Code. I spec, orchestrate, and run the evals.
Knowing what to stop
Built a Slay the Spire 2 stats site to 250 users in six weeks, then killed it on a cost call. Knowing what not to keep alive is the rarer skill.
I trained the models first
Through Pareto: prompt engineering and adversarial testing, among the highest-rated of 2,000+, on projects for companies like Anthropic and Stanford.

A conviction bet, in order.

2023 · the AI turn
The turn
A side gig that rerouted everything.
While managing a team and studying business admin, I started training AI through DataAnnotation after hours. My first real look at how the frontier models get built.
Maps to

I moved on frontier-lab AI training before it was a category. Unburdn sells that same early-mover instinct to its clients.

~April 2024 · the bet
The bet
Invited to Pareto. I paused college to take it.
Paused the associates a semester short to chase execution over the credential while the window was open. I may finish it. The bet was right.
Maps to

Speed is strategy, and I play for first-mover advantage. Unburdn runs on the same clock: move before the market is ready.

~October 2024 · regulated world
The hard room
Production constraints, the unforgiving kind.
Joined a finishing shop as an estimator and built the first calculator. Offline, controlled, zero room for a demo that falls over.
Maps to

I'm at home where stakes are real, and I understand how regulation shapes what you can actually ship. Those are Unburdn's hardest rooms.

2025 · senior work + prototype
Build mode
From grading models to shipping them.
Through Pareto, senior adversarial work on projects for companies like Anthropic and Stanford. Then I started building with Claude, built the MFC prototype, and offered the shop a $10K license.
Maps to

I crossed from evaluating AI to deploying it for a paying customer. That crossing is the whole job at Unburdn.

Early Jan 2026 · it landed
It landed
The tool sold. I got promoted, not cut.
The license closed. The estimating department went from two to one (~$50K/yr salary-only), and the shop kept me on as Business Systems Developer.
Maps to

Production AI, a measured outcome, a human in the loop. By your ladder, that is Run.

Now
Now
I found the company built for this.
Unburdn's whole thesis is getting AI adopted where it's resisted. The arc points straight here.
Maps to

Adoption where it's hard isn't a pivot for me. It's been the assignment all along.

MOD Pizza training manager (reviews up 20%+ in six months). Founder of Empty Ego ($500 to six figures over 22 months). Sales and member services at Sunrun and UnitedHealth.

Before the calculator: Excel, HTML, docs, ERP add-ons. Then the realization that an offline Electron app was the right shape, which became MFC.

An event-inventory tool and a site-builder. Formed Psyrcuit LLC around Nov 2025. Productized MFC, and built the stats site I later killed.

The shop sold to a PE owner. Through them, a second finishing company is in discussion for the same $10K structure. On the table, not closed.

By your own ladder, I'm already at Run.

Crawl · Walk · Run

Run is agentic AI in production, with a human in the loop, in a real environment. That's what you just poked at. I'm not telling you I could get there. I'm there.

All of it is pokeable on purpose. If a claim didn't survive you pushing on it, tell me. That's the standard I hold my own work to.

I'm Chase Lance. I ship production AI where it's hard to ship. If that's who you want in the room, I'd like to be in yours.

See you Tuesday.